Relationship between drainage capacity and foundation settlement at Pre Rup Temple: Basic study on Khmer architecture drainage systems Part 1

Abstract

In the ancient city of Angkor, heavy rainfall was common in the rainy season and rapid rainwater drainage was important for maintaining urban functions. This study focuses on the drainage system of an example of Khmer architecture from the Angkor period, the Pre Rup temple built in the year 961. We analyze the relationship between the drainage capacity and foundation settlement, revealing that the foundation settlement was larger near drainage channels with a low drainage capacity. After showing that only a few of the drainage channels function properly, it is possible to have some certainty on the relationship between the drainage capacity and foundation settlement.

Introduction

The ancient city of Angkor in Cambodia, located in the modern province of Siem Reap, experiences severe downpours during the rainy season. Efficient drainage of rainwater is essential for maintaining urban functions. Most of the religious architecture of the Khmer era, which centered on Angkor, was built on a foundation of rammed earth laid with stone, leading to repeated debate as to whether foundation settlement due to poor rainwater drainage has been a major factor in the collapse of the buildings.^1,2

The authors have been examining³ the 10th-century Hindu temple complex of Pre Rup.,^4–6[Note 1)] There are scattered foundation irregularities at the lower level of the temple and its brick towers have a tilt, which is particularly serious in the case of the southeast tower. An Italian team has recently been monitoring the displacement behavior of this tower, and has carried out strength tests of the foundation and superstructure. Consequently, it was revealed that the main factor causing the displacement of the tower’s foundation was low proof strength; therefore, structural reinforcement of the foundation was carried out.^7,8 It was also pointed out that the factors influencing the foundation settlement are fluctuations in the groundwater level and rainwater stagnation, but the causal relationship was not clarified.

The impact of failure to ensure proper drainage of rainwater may not become apparent in the short term, but is important for the long-term conservation of buildings. This study focuses on the rainwater drainage systems used in Khmer religious architecture and evaluates the effect of rainwater drainage on the buildings. The ultimate purpose of this study is to demonstrate the importance of paying attention to such technical issues for the preservation of the buildings in the future.

Rainwater drainage studies began with the evaluation of the drainage channels at Bakong and Pre Rup by École Française d'Extrême-Orient.^9,10 The drainage method used at Bayon temple, built at the end of the 12th century, was discussed by Jacques Dumarcay.¹¹ In recent years, So Sokuntheary reported on the drainage at the Bayon temple complex under the Japanese Government Team for Safeguarding Angkor (JSA) restoration and maintenance plan.¹² This study makes an important contribution that quantitatively assesses the present situation from field survey data, demonstrating that most rainwater permeates into the foundation. Furthermore, the authors categorized the various drainage channels and clarified the drainage system used in each era prior to the 13th century.¹³ They pointed out that dysfunctional drainage channels are the cause of building collapse, but the exact relationship and overall drainage method of the temple complex have not been fully verified.

It is clear that there has been insufficient research on how rainwater drained from the temple complex up to the 13th century, thus, the data on each individual temple is limited. Considering this, a significant issue is to understand the basic relationship between rainwater drainage characteristics and the present state of each temple.

It is for this reason that in this work we take up the case of the abovementioned Pre Rup temple and aim to clarify the relationship between the rainwater drainage characteristics and its present state. The temple was built in the year 961 and since then, channels designed to drain rainwater from the site have been damaged or displaced, or are filled with accumulated sediment. To evaluate the characteristics of rainwater drainage at the temple, the evaluation focuses on the drainage capacity of the system as it was constructed. We first obtain the measurements[Note 2)] of each drainage channel to obtain its maximum flow capacity then examine the rainfall data to determine the amount of rainwater runoff each would receive. Finally, we compare these results to obtain their drainage capacity.

Because the superstructure of each building in the temple is made of different materials and structures and is in various states, the displacement characteristics cannot be verified and reliably compared against each other. Furthermore, displacement data for each building are not available. However, the paving foundation has a similar composition throughout the temple. In this situation, we do not consider the superstructure of each building in this study, but focus on the subsidence of the paving foundation. The foundation subsidence is obtained as the level deviation of the paving top surface at each location, as obtained from survey data,[Note 3)] with reference also to past restoration records.

Based on the determined drainage capacity of each drainage channel and the verified foundation subsidence, we examine the relationship between the two, and consider the characteristics of rainwater drainage at the Pre Rup temple and its effect on the foundations. Based on these results, we propose possible measures to prevent further subsidence.

Part of this paper was presented at the 2018 Architectural Institute of Japan Conference. This version has been significantly revised and modified.

Location and Profile of Each Drainage Channel Basic structure of drainage channels

Pre Rup is a temple of the composite pyramid type in which the buildings are constructed on stacked foundations, referred to as terrace foundations in this work. The temple is divided into three zones (Figures 1 and 2).[Note 4)] The first and lowest zone and the second zone above it consist of a single-tiered terrace foundation and each zone is surrounded by a wall. The three-tiered terrace foundations forming the third and upper zone have no surrounding wall; thus, rainwater is not collected but flows out all around to the lower zones. Rain that falls in the second zone with that flowing from the upper zone is gathered into various drainage gutters (ditches) that gather the water into a channel at the base of the second surrounding wall from where it is discharged into the lower first zone. Similarly, in the first zone, rainwater is gathered by various gutters and discharged outside the temple via a drainage channel at the base of the first surrounding wall. Thus, there are drainage gutters throughout the temple that channel rainwater out of the temple to various locations. Because this paper focuses on the drainage capacity of the main drainage channels, we leave the examination of the full drainage plan of the temple, including the individual drainage gutters, to the next paper. We analyze the drainage channels at the base of the first and second surrounding walls that lead the runoff water out through a pipe structure in the wall.

FIGURE 1. East elevation and section of Pre Rup templeFigure 1 is drawn by the authors from Reference 19. All others were created by the authors from fieldwork.

FIGURE 2. Site plan of drainage channel at Pre Rup temple

Each side of each temple zone is divided into two sections and, as shown by So Sokuntheary,¹³ there are 16 drainage channels in total, one in each section. In this work, the drainage channel at the southern end of the east side is labeled as “SE” and marked with ① in Figure 2. The other channels are numbered clockwise around the temple. Each drainage channel (Figures 3-5) consists of a culvert running within the surrounding wall, an inlet hole set into the pavement of the terrace foundation and an outlet on the sidewall. The inlet hole and outlet are aligned perpendicular to the wall, and the distance between them is almost equal in every case. The outlet protrudes from the wall and consists of a sandstone block of approximately 50 square cm. According to So Sokuntheary,¹³ this block is decorated with the head of the Macara, an imaginary creature symbolizing water. The mouth of the Macara is a round or a rounded square hole that forms the exit of the pipe structure.

FIGURE 3. Section of each drainage channel in zone 1

FIGURE 4. Section of each drainage channel in zone 2

FIGURE 5. Elevation of drainage channel

This pipe structure is part of a culvert cut approximately 20 cm into the top of the sandstone block, which in turn links to the culvert cut into the laterite terrace that draws water from the inlet hole. From the structure of interlocking components, it is clear that the drainage channels were constructed as part of each terrace foundation.

Additionally, the sandstone block with the outlet hole protrudes from the terrace foundation leaving approximately 60 cm set into the foundation, where it is held in place by the wall above it. This indicates the structure cannot be replaced without removing the surrounding wall. (Although in the case of drainage channel ① there is discontinuous stone masonry in the wall above the drainage channel.) Furthermore, the Macara sculpture carved as part of the outlet hole has the same form in each zone.

From these observations, although there are signs of repairs to each outlet, it seems that these drainage systems were installed during the construction of the surrounding wall for each zone.

Location and current status of drainage channels

The drainage channels in the first zone are all located approximately equidistant from a corner of the surrounding wall at a distance of 15 to 17 m (Figure 2). This cannot be confirmed for drainage channel ③ because it is buried under the stone blocks of the collapsed surrounding wall. Furthermore, the culvert at drainage channel ① cannot be surveyed owing to a reinforced concrete structure that has been constructed at the front. Moreover, the inlet holes of drainage channels ④ and ⑧, and the outlet holes of drainage channels ① and ⑧ were found to be buried in sediment.

In the second zone, there are buildings attached to the surrounding wall, thus, the drainage channel positions are different on each side (Figure 2). Drainage channels ⑨, ⑫, ⑬, and ⑯ on the east and west sides, and drainage channel ⑮ in the northeast, are located between two attached buildings. Drainage channels ⑩ and ⑪ on the south side, and drainage channel ⑭ in the northwest are located almost at the center point of one long attached building. Among these drainage channels, ⑨, ⑪ and ⑯ were found to have a large amount of sediment and dust accumulated in the culvert.

No rainwater drainage was observed through the drainage channels where sediment had accumulated, even during heavy rainfall. We first cleaned up all the drainage channels and then measured each culvert using a laser distance meter. We begin with an analysis of the profile of the drainage channels in the second zone, which is nearer the center of the temple.

Profile of drainage channels

We analyze the shape of the drainage channels by focusing on the slope of the culvert in the terrace foundation between the inlet hole and outlet, and the slope and cross-sectional area of the pipe structure at the outlet.

The position of the front edge of the inlet hole of each drainage channel is marked A. The position of the innermost end of the pipe structure at the outlet is marked A'. The inflection points of the gradient of the culvert and pipe structure are marked with sequential letters (Figures 3 and 4). Table 1 lists the gradients between these points and Table 2 presents the cross-sectional area at each point in the outlet pipe structure. As will be described later (Chapter 4), there is observable subsidence near the drainage channels in the surrounding wall of the first zone. Of this, the subsidence in the orthogonal direction from the surrounding wall is observed only in drainage channels ⑦ and ⑧.

TABLE 1 Gradients between points from the inlet hole of each drainage channel to the outlet

		A-B	B-C	C-D	D-E	E-F	D/E/F-A'	A'-B'	B'-C'	C'-D'	D'-E'	A'-B'/C'/D'/E '
Zone 1	①	×	×	×	×	×	×	−0.178	-	-	-	−0.178
	②	−0.080	−0.007	0.156	0.014	-	−0.028	−0.618	−0.353	−0.266	-	−0.359
	③	×	×	×	×	×	×	×	×	×	×	×
	④	−0.007	-	-	-	-	-	−0.098	-	-	-	−0.098
	⑤	−18.571	−0.133	0.090	-	-	−0.013	−0.356	−0.217	0.147	-	−0.065
	⑥	−0.002	−0.167	0.022	-	-	0.173	0.173	−0.054	0.101	-	0.078
	⑦	−14.000	−0.134	0.358	−0.385	0.238	−0.096	−0.330	0.044	−0.319	-	−0.235
	⑧	−4.070	−0.136	0.198	-	-	−0.012	−0.033	−0.202	−0.077	-	−0.118
Zone 2	⑨	−17.200	−0.200	−0.035	-	-	0.115	−0.872	−0.375	−0.087	-	−0.263
	⑩	−0.887	−0.022	−0.048	0.012	−0.065	−0.012	−0.443	−0.175	−0.382	-	−0.294
	⑪	−0.035	−0.075	−0.027	−0.176	-	−0.047	−0.357	−0.524	−0.219	−0.211	−0.337
	⑫	−0.053	−0.095	−24.087	−0.045	0.013	−0.121	−0.163	−0.308	−0.127	0.088	−0.095
	⑬	−0.029	−0.245	−0.008	-	-	−0.046	−0.062	×	-	-	−0.059
	⑭	−0.001	−18.693	−0.282	−0.007	−0.068	0.048	−1.000	−0.103	−0.221	−0.071	−0.155
	⑮	−30.264	−0.046	0.042	−0.066	-	0.118	−0.067	0.075	-	-	−0.005
	⑯	−2.096	−0.157	0.019	−0.057	0.004	−0.075	−0.461	−0.130	−0.040	-	−0.176

TABLE 2 Cross-sectional area at each point in the outlet pipe structure

(cm2)		A'	B'	C'	D'	E'
Zone 1	①	-	132	-	-	-
	②	30	50	120	180	-
	③	×	×	×	×	×
	④	-	95	-	-	-
	⑤	80	110	160	150	-
	⑥	120	110	110	120	-
	⑦	28	51	55	86	-
	⑧	10	10	60	80	-
Zone 2	⑨	10	50	70	130	-
	⑩	50	40	80	100	-
	⑪	10	20	70	80	×
	⑫	70	90	140	150	140
	⑬	140	110	×	×	×
	⑭	50	80	90	140	170
	⑮	100	130	200	-	-
	⑯	30	90	90	150	-

From this, it is considered that the drainage channels that extend outward from the surrounding wall at a right angle still maintain the slope they had at the time of construction, except in the case of channels ⑦ and ⑧.

Profile of drainage channels in the second zone

In the second zone drainage system, the Macara sculpture forms the outlet at the mid-height of the terrace foundation sidewall (Figure 5d), and there is a large difference in the level between the inlet hole and outlet (Figure 4). The profile of the drainage channel differs between the eastern half (front) of the temple and the western half (rear).

On the eastern half of the second zone, the inlets consist of a rectangular hole (Figure 5a) cut into the surface of the terrace pavement (Figure 5a). Between the low point B of each inlet hole and the innermost part A' of the outlet pipe structure, the culvert slopes gently downward at approximately 5%. The outlet pipe structure slopes steeply downward at more than 10%. At the outlet, the cross-sectional area of the spout is 100-200 cm², whereas at the innermost end A', it is less than 50 cm², much smaller than at the spout, except in the case of ⑮. Water passing through the pipe structure with this narrowed inner section rushes out in a parabolic curve because the flow is uniform with constant velocity, pressure, and density.[Note 5)] This behavior was observed in the rainy season. Furthermore, the Macara sculpture at the outlet is delicately decorated with teeth in the jaw, a tongue, and curly hair on the cheeks. It seems that here, on the temple frontage that is easily seen by people (the eastern half), the drainage system was planned to ensure that a parabolic flow of water would be observed coming from the mouth of the beautiful Macara.

The inlets on the western half are different. Here, they consist of a gap formed by leaving out one laterite block from the foundation of the surrounding wall (Figure 5b) and it can be inferred that the feature has a simple design compared with those on the eastern half.

Between inlet hole A and the innermost part A' of the outlet pipe structure, there is a steep downslope of approximately 25% (drainage channel ⑬) or a step (drainage channels ⑫ and ⑭), whereas the rest of the structure is gently sloped. This gentle downslope as well as the cross-sectional area at the innermost part A' is more than 70 cm² except in the case of drainage channel ⑪, larger than that on the eastern half. No parabolic curve of discharge water is observed there.

Profile of drainage channels in the first zone

The drainage channels in the first zone are located higher up on the foundation (Figure 5c), with the Macara sculpture at the outlet almost at the same height as the inlet. The profile of the culverts is quite different from that of the second zone.

Each inlet hole passing through the surrounding wall foundation is different in the first zone (Figures 5b and 6); none has a rectangular shape as in the second zone. Although the drainage channels are unfinished and rather uneven in the first zone, we measured a downward slope measuring approximately 15% between the inlet hole and a position 50-100 cm toward the outlet, and thereafter a rising gradient of approximately 20% from there to the innermost part A' of the outlet pipe structure. Thus, rainwater remains in the culvert until it reaches the high point of the profile at A'. It is inferred that this upslope in the culvert is a feature for precipitating solid matter from the rainwater by retaining it for a while. In one case, drainage channel ④, the culvert does not pass through the wall but stops approximately 50 cm from the inlet hole, whereas other channels are roughly constructed; it appears that plans changed during the construction of the drainage system.

FIGURE 6. Inlet hole of drainage channel

The outlet pipe structures on the eastern half are steeper than those on the west, and the cross-sectional area of the innermost part A' is less than 30 cm², less than that on the western half. As with the second zone, the cross-sectional shape of the outlet pipe structures is different on the east than on the west. However, all Macara heads at the outlets from the first zone are more roughly sculpted than those of the second zone; they have a formal round mouth, little decoration, and asymmetrical eyes. The culvert in drainage channel ⑥ runs uphill from the inlet to the outlet. Drainage channel ④ has an outlet pipe that penetrates only 15 cm in from the spout. In drainage channel ①, the back of the outlet pipe structure does not penetrate the wall, leaving a blockage of a few centimeters, thus, the water path is closed off. The reason for these flow interruptions in the drainage channels is not clear. One possibility is that construction came to an end. Another possibility is that Macara sculptures were installed higher, where the work would have been easier, rather than at the foundation mid-height by masons who did not understand the drainage plan. This would also explain the cases where the inlets and outlets are not connected. It is also possible that the drainage plan was changed during construction; rainwater harvesting may have been adopted in the first zone at a certain stage.

It is likely that the towers and storehouses in the first zone were constructed by the king who succeeded to the throne after Rajendravalman, who built Pre Rup. Therefore, there is room for argument as to whether the drainage channel plan underwent changes at that time. However, any examination of whether the drainage channel arrangement changed would need to be done when checking the overall temple layout. Therefore, in this study, we only point out the possibility; it will be discussed later in another paper. Regarding the construction of the drainage channels in the first zone, no further clarification is possible in this paper; however, it is clear that the observed drainage plan promotes the retention of rainwater.

Chapter conclusion

The 16 drainage channels found in the Pre Rup temple have different cross-sectional shapes depending on their location. The main differences observed in the drainage plans are as follows.

At the front of the temple, which faces east, the outlets of the drainage channels have a narrow cross section and rainwater is discharged in a parabolic arc from the delicately decorated mouths of the Macara sculptures. It is clear that drainage was planned in consideration of how water would flow away. However, at the rear of the temple, which faces west, the outlet pipes have a larger cross section and no parabolic discharge is observed. Because the drainage channels at the rear of the temple are less visible, it can be assumed that the purpose was to quickly discharge as much rainwater as possible.

Regarding the drainage channels in the second zone, where the outlets are lower than the inlets, the culvert in the terrace foundation has a consistent downslope and it seems that a plan was devised to promote drainage. The drainage channels in the first zone have outlets at almost the same height as the inlets and the culvert has a low point near the middle of the channel, forming a feature that promotes the retention of water. In the first zone, drainage work appears to have been interrupted, with some drainage channels that do not penetrate, leaving a barrier of several cm. These details suggest that the retention of rainwater became acceptable owing to some change in the drainage plans. With these different drainage channel profiles in the first and second zones and between the eastern and western sections of the temple, the flow velocities and flow rates are expected to differ along with the visible shape of the water discharge. In the next chapter, we examine the drainage capacity of each of the evaluated drainage channels.

Drainage Capacity of Each Drainage Channel Method of examining drainage capacity

The profile of the pipe structure of the channel is the main determinant of whether rainwater entering the drainage channel flows smoothly to the outlet spout or gets backed up in the channel. Measurement data for each outlet pipe are used to calculate the maximum flow capacity through each drainage channel. As described in Chapter 2, at the outlet of each drainage channel, the slope of the pipe structure and the cross section are not constant. To perform the calculations, the inside of the pipe structure is regarded as a composite pipeline consisting of multiple connected pipe sections with different slopes and cross sections. This composite is then replaced by a single conduit of a constant slope and profile that maintains the hydraulic equivalence. Using the Hazen–Williams equation, which is the most general formula for average flow velocity in pipes, the average flow velocity and flow rate in this equivalent single pipe area are calculated. From the results of these calculations, we examine the drainage capacity.

The Hazen–Williams equation is expressed by Equation (1):[Image Omitted. See PDF]where v is the velocity [m/s], C is the roughness coefficient, R is the hydraulic radius [m], and I is the slope of the energy line. The hydraulic radius R is a value obtained by dividing the cross-sectional area of flow A by the length of the line described by the intersection of the channel wetted surface with a cross-sectional plane normal to the flow direction (wetted perimeter, p), and is given by Equation (2):[Image Omitted. See PDF]where R is the hydraulic radius [m], A is the cross-sectional area of flow, and p is the wetted perimeter. If drainage is through a circular pipe of diameter d [m] in a full state,[Image Omitted. See PDF]therefore, R in this case is:[Image Omitted. See PDF]

By substituting this relationship into Equation (1), Equation (3) is obtained as the relational expression in the case of a circular pipe:[Image Omitted. See PDF]where, if flow rate Q [m³/s] is used as a hydraulic element instead of the average flow velocity v [m/s],[Image Omitted. See PDF]

From this relationship, equation (4) is obtained instead of equation (3):[Image Omitted. See PDF]

In this study, Q is the maximum flow capacity of rainwater at each outlet, and if Q is larger than the amount of rainwater entering the drainage channel even during heavy rain, it is inferred that the outlet has the ability to properly drain the water. In the opposite case, it is deduced that rainwater may back up and cause flooding in that zone.

Examination of drainage capacity of each drainage channel

The process of determining the drainage capacity is explained using outlet ⑩ (Figure 4) connected to drainage channel ⑩ in the second zone.

First, each section (A′B′, B′C′, and C′D′) is converted into an equivalent circular pipe whose diameter is the short side of the cross section. The pipe structure of the outlet can then be regarded as a compound pipe consisting of three connected circular pipes as follows:[Image Omitted. See PDF]where l is the pipe length [m]. Roughness coefficient C, which indicates the roughness of the inner surface of the pipe, differs depending on the material of the pipe and is indicated by a numerical value of 60-150.¹⁴ The larger the value, the smoother the pipe. There is no available data on the C value corresponding to sandstone, which is the material of the outlet. We choose a C value of 60, representing the highest resistance, to obtain the drainage capacity under the most severe conditions.

Next, to replace the compound pipe with one single pipe, the pipe diameters d_A′B′ of section A′B′ (0.07 m) and d_C’D’ of section C′D′ (0.09 m) are converted to a pipe diameter of d_B′C′ (0.06 m), which is the minimum diameter of the three-section compound pipe, using equivalent line lengths l' as calculated using equation (5).[Image Omitted. See PDF]

However, d₁ > d₂. From equation (5), $l_{A^{\prime }{B}^{\prime }}^{\prime }$ and $l_{C^{\mathit{\prime }}{D}^{\mathit{\prime }}}^{\prime }$ are.[Image Omitted. See PDF][Image Omitted. See PDF]

Therefore, the compound pipe between A' and D' can be replaced with a series of the following three single pipes.[Image Omitted. See PDF]

As a result, the compound pipe between A' and D' is equivalent to the single pipe described as follows:[Image Omitted. See PDF][Image Omitted. See PDF]

Because the height of section A'D' is 0.089 m, the slope of this single line is.[Image Omitted. See PDF]

Therefore, the maximum flow capacity Q_⑩ of rainwater at outlet ⑩ can be calculated using equation (4).[Image Omitted. See PDF]

During heavy rains in this location, the rainfall intensity reaches 75.4 ㎜/hr,¹⁵ thus, the rainwater runoff per drainage channel in the second zone is 0.011 m³/s.^16,17 Comparing this with Q_⑩, the amount of rainwater outflow > the maximum flow capacity, thus, it is theoretically deduced that rainwater would not be properly drained from outlet ⑩ and would back up in the second zone.

Table 3 lists the maximum flow capacities of each outlet calculated by this method, and Table 4 presents the amount of rainwater runoff. In the second zone, only two outlets (⑫ and ⑭) have sufficient capacity to properly drain rainwater during heavy rains.

TABLE 3 Maximum flow capacity of each outlet (m³/s)

Zone 1		Zone 2
①	0	⑨	0.004
②	0.005	⑩	0.006
③	Unmeasurable	⑪	0.001
④	0	⑫	0.017 ※
⑤	0.015	⑬	0.009
⑥	0	⑭	0.013 ※
⑦	0.006	⑮	0.006
⑧	0.001	⑯	0.003

※ The maximum flow capacity > The amount of rainwater

TABLE 4 Amount of rainwater outflow (m³/s)

Zone 1	Zone 2
0.03	0.011

Table 5 lists the estimated amount of rainwater backup in the first and second zones during heavy rain. These estimated volumes can be calculated by subtracting the maximum flow capacity (Table 3) of each outlet from the amount of rainwater runoff per drainage channel (Table 4). Accordingly, the flood volume is approximately 133 m³/h in the second zone and approximately 769 m³/h in the first zone, a significantly larger figure than that in the second zone. This is because the amount of rainwater received by the first zone is greater than that which enters the second zone. It is also affected by the existence of outlets whose slope is reversed and that do not penetrate into the first zone (i.e., those with an inlet lower than the outlet, and where the maximum flow capacity is set to 0).

TABLE 5 Estimated amount of rainwater (m³/h)

Zone 1	Zone 2
768.82	132.475

Chapter conclusion

The maximum flow capacity of each drainage channel outlet is calculated from its gradient profile and cross sectional shape. This is compared with the amount of rainwater runoff during heavy rain. The main findings of this evaluation are as follows.

Except in some cases, the maximum flow capacities of the drainage channels are greater on the western half (rear) of the temple. Two drainage channels on the western half of the second (inner) zone have the capacity to drain all the rainwater runoff, even during heavy rain. It appears that the plan was to simply ensure the discharge of more rainwater on this side, in contrast with the conscious effort to achieve smooth parabolic discharges from the outlet spouts as observed on the front (eastern half) of the temple.

Considering the drainage capacity of each zone as a whole, both appear to be prone to flooding during heavy rains, but the amount of backed-up rainwater runoff per unit time is considerably larger in the first zone. This is due to the inflow from the higher zones and some of the outlets do not drain any water because they fail to penetrate the surrounding wall. It is clear that the two zones have very different drainage capacities, thus, the amount of time during which each zone remains flooded during heavy rain is different. This is expected to have an effect on the subsidence of the foundations. In the next chapter, we examine the foundation settlement in each zone.

Foundation Settlement

Various restoration and conservation work has been carried out in different parts of the Pre Rup temple.¹⁸ Therefore, we first conduct a literature evaluation to determine if any foundation repairs have been carried out,[Note 6)] and then examine the foundation settlement considering the findings.

Foundation repairs recorded in the literature

The first records of maintenance and restoration at the Pre Rup temple appear in the 1930 Annual Report of École Française d'Extrême-Orient. At that time, the temple was buried in earth and sand up to near the top of the pyramid, and the site was overgrown with vegetation. By about 1935, the removal of the sediment was almost complete.[Note 7)] Comparing photographs taken immediately after removal with the current situation, it is observed that some structures have disappeared, but the overall composition of the temple has been retained. Records show that almost all the individual buildings have undergone partial superstructure repairs. However, none of the repairs have involved complete dismantling of the foundation. The paving stones forming the terrace foundation have been re-laid with curbs and otoliths around the edges, but the central paved area has not been completely dismantled and replaced.

Method of analyzing foundation settlement

We examine the foundation settlement by obtaining level data for the paving stones and analyzing variations in the amount of settlement at certain locations. The terrace foundation was marked out with 2 m grid lines in the east–west and north–south directions and the paving stone level was measured at each grid intersection and at each point of contact between a grid line and a structure. For the individual building foundations, we obtained measurements around the outer perimeter of the foundation and in the vicinity of the building wall. Measurements were taken at each inside and outside corner as well as at intervals of approximately 0.5 m between the corners. For the positional relationship of each part of the temple, we use survey data presented in a previous paper.³ The terrace foundation of both the first and second zones incorporates multiple steps, and some buildings of the same type have foundations with different heights. We classify the level data for these different-height foundations separately, taking the highest point among foundations of the same height as the datum. Furthermore, for each foundation, we calculate the deviation at each measurement point from the average of all points. In the figures, we indicate the level distribution by plotting six gradations (darker is lower) with one gradation equivalent to 50 ㎜ (Figures 7-12). These plots are used to examine the foundation settlement based on the level distribution of each foundation.

FIGURE 7. Level distribution of terrace foundation in Zone 3

FIGURE 8. Level distribution of building foundation in Zone 3

FIGURE 9. Level distribution of terrace foundation in Zone 2

FIGURE 10. Level distribution of building foundation in Zone 2

FIGURE 11. Level distribution of terrace foundation in Zone 1

FIGURE 12. Level distribution of building foundation in Zone 1

Foundation settlement in each zone Third zone

In the third zone, the terrace foundation consists three steps. At the top step, the level of each paving stone adjoining the building foundation of the central main tower (A1) is higher than that of the others (area c in Figure 7) and there is no significant difference on the four sides of the tower. That is, there appears to be little paving stone subsidence in the center of the terrace foundation. On each side, the paving level falls off from the center toward the outer edge; this is recognized as a water-shedding gradient. However, the curb has locally subsided in the center of each side and at the corners. The building foundations of the sub-towers (A2 to A5) are also low near the corners of the terrace foundation (Figure 8) and this subsidence seems to correspond with subsidence of the terrace foundation. On the second step of the terrace foundation, the paving stones have a water-shedding gradient in places, but the corner stones of the curb have subsided (area b in Figure 7). Further, the southeastern corner has collapsed. On the first step of the terrace foundation, the paving stones are low around the middle of each side and subsidence at the corners is particularly large (area a in Figure 7). The foundations of the small towers (B1 to B12) on each side are also low near the center of the terrace foundation (Figure 8) and this seems to correspond with subsidence of the terrace foundation.

Second zone

The second zone consists of a terrace foundation with a single step, and there is a lower gutter between the surrounding wall and the storehouses (C1 to C12) (area a in Figure 9). The paving stones of the terrace foundation are high around the center (area b in Figure 9). The paving stone level is different on each side, but on all sides, the level falls from the center toward the outer edge, thus, a water-shedding gradient is recognized. The footings of the upper (third zone) terrace foundation are not markedly uneven, but the building foundations of the storehouses located on each side have subsided on the side with the drainage channel (Figure 10).

First zone

The first zone also has a one-step terrace foundation and the surface of the paving stones on the terrace forms the main level. On the north, south, and west sides, there are two additional step terraces adjoining the upper (second zone) terrace foundation. We refer to these as the main terrace and additional terraces in this work. The paving stones of the main terrace are very uneven in appearance and the levels are quite different on each side. On the eastern half of the north side “EN,” the paving stone level is high toward the center of the temple and low toward the outer edge, thus, a water-shedding gradient is observed (area a in Figure 11). There is little unevenness here. On the western half of the south side “WS,” the paving stone level is low in many places, including toward the center of the temple, and no general tendency is observed nor is there a water-shedding gradient.

On the upper additional terrace, the level of the paving stones is higher in area “EN” than the other side (areas b and c in Figure 11), and the level of the foundations of storehouses E7 and E8 above the additional terrace is not uneven in area “EN” (Figure 12). In areas other than "EN," the storehouse foundations (E1 to E5) have subsided at the outer perimeter of the temple; that is, the level of the storehouse foundation seems to have subsided in conjunction with the terrace foundation below it. Subsidence is particularly large in area “WS,” where the surrounding wall foundation has also subsided significantly and part of the surrounding wall has collapsed.

On the eastern side, where there is no additional terrace, there are three towers (F1 to F3) on the high foundations at area “SE” and two towers (F4 and F5) at area “NE.” The foundation of tower F1 located beside the east gate has less displacement than the other towers. However, the foundation of tower F3 at the south end of area “SE” has subsided by 20 cm or more at its southeast corner and that of tower F4 at the north end of area “NE” has subsided by the same amount at the northeast corner. It should be noted that the state of the tower foundations on the east side is almost axisymmetric with respect to the east-west centerline, with greater subsidence further from the centerline. Because there are drainage channels on the perimeter of the foundation where there is large subsidence, it is considered that this subsidence is related to the rainwater drainage.

Chapter conclusion

Measurements of the foundation settlement indicate differences among the zones, as follows.

In the third zone, the central paving stones on the third terrace foundation have subsided less and there is a water-shedding gradient, but the curbs have locally subsided at the center of each side and at the corners. On the first step of the terrace foundation, the paving stones have subsided in the middle of each side. All subsidence of the building foundations is observed to correspond with the subsidence of the terrace foundation. In the second zone, the paving stone level adjoining the upper (second zone) terrace foundation is different on each side, but on all sides, there is a water-shedding gradient from the center outward. The footings of the upper terrace foundation have less unevenness, but the storehouse foundations have subsided on the drainage channel side. In the first zone, there is a water-shedding gradient on the northeast terrace foundation, and there is less unevenness here than on the other sides. However, the southwestern terrace foundation has a large amount of subsidence overall, and a part of the surrounding wall above it has collapsed. On the east side, the foundations of the south tower and north tower have a large amount of subsidence almost axisymmetric with respect to the east-west centerline, and there is a drainage channel on the perimeter here. There appears to be a relationship between the foundation subsidence and rainwater drainage. In the next chapter, we examine the relationship between the foundation subsidence and the drainage capacity of each drainage channel.

Relationship Between Drainage Capacity and Foundation Settlement in Each Zone

We examine the relationship between the drainage capacity of the drainage channel and the foundation settlement in each zone including the third zone, which is the center of the temple, with the results of observing water pooling during heavy rain.

Third zone

In the third zone, there is little subsidence of the paving stones in the central part of the uppermost terrace foundation, and a water-shedding gradient is observed in all directions. There were no areas found to become submerged even during heavy rain, and rainwater runoff quickly flowed out to the lower terraces. Here, settlement would be anticipated because the terrace foundation supports the load of the central tower group and it is important to prevent the permeation of water into the foundation in this area. There seems to have been a plan to ensure rainwater promptly drained via a water-shedding gradient in all directions, and there is no surrounding wall. However, the curbstones around the terrace foundation have local subsidence in the middle of each side and at the corners. The curb has been repaired, but there are indentations on the top surface of the curbstones in the middle areas. Because there is a staircase in the center, it seems that these stones were locally worn by the passage of worshipers. However, the corner subsidence of the curb is thought to be due to the outward opening of the foundation, a mechanical and structural phenomenon, because the joints between the sandstone blocks in these areas have spread apart.

The paving stones of the first terrace foundation in this zone show subsidence in the central area of each side. There is no opening of the joints or deterioration caused by plant roots. Because this central part supports the upper terrace foundation, this subsidence is regarded as evidence of settlement. The small tower foundation has also subsided significantly at the center of the temple, corresponding with the subsidence of the terrace foundation. During heavy rains, water ponds at the center of the foundation where there is settlement. It is apparent that a large amount of this rainwater infiltrates into the foundation, leading to fear of a major collapse of the foundation and the small tower in the future. Immediate measures should be taken to prevent rainwater from penetrating into the foundation.

Second zone

In the second zone, there is no major unevenness in the paving stones of the terrace foundation and the footings of the upper terrace foundation. The eight drainage channels arranged around the perimeter have a consistent downslope and are shaped to promote drainage. The two drainage channels on the west side have sufficient drainage capacity, but given the overall drainage capacity of the channels, it is expected that heavy rains will cause temporary flooding, although flooding will last only a short time. During actual heavy rain, no water retention was observed except in areas where the stone was damaged. There was no pooling in the center of the terrace, and rainwater runoff instead collected in areas like the lower gutter around the surrounding wall.

The upper (third zone) terrace foundation stands on an independent foundation in the center of the second zone. Therefore, the paving stones of the foundation do not support a large load and settlement is not expected.[Note 8)] It appears that the temple drainage plan assumed that water would be cleanly discharged through the mouths of the Macara sculptures on the front side of the temple facing east, whereas any excess rainwater would accumulate on the outer periphery.

First zone

The foundation settlement in the first zone varies by location. Further, the drainage capacity of the drainage channels arranged around the periphery also differs greatly. In area “EN” there is almost no unevenness of the terrace foundation, the maximum flow capacity of drainage channel ⑦ is 0.006 m³/s, and it seems that the drainage is adequate although rainwater is retained temporarily. During actual heavy rain, the retention of water was confirmed only in one area along the surrounding wall. However, subsidence of the terrace foundation is significant in area “ES” and drainage channel ④ is blocked. There is no drainage for runoff water to flow, thus, water is retained in the vicinity for a longer period causing it to penetrate into the foundation. During actual heavy rain, it was observed that the entire area was flooded with rainwater for a long time. Of the five eastern towers, the south and north towers have subsided by approximately 20 cm in the southeastern corner of the south tower and the northeastern corner of the north tower. In front of each of these subsided areas are drainage channels: ① where the pipe structure does not penetrate the surrounding wall and ⑧ with an extremely low maximum flow capacity of 0.001 m³/s. In this area, it was also observed that rainwater ponds for a long time during actual heavy rain.

These eastern towers are thought to have been expanded after initial construction work, increasing the settlement load on the foundation beyond that expected at the time of construction. The drainage channel on the front (east side) of the temple has a constricted pipe structure and it creates a good flow that produces a parabolic outlet stream, similar to the drainage channels on the east side of the second zone. However, in the first zone, the culverts leading to the drainage channel outlets all have profiles that cause water to be retained and some of the drainage channels do not penetrate the terrace foundation. It is probable that these features resulted from changes in the drainage channel plan. Because understanding this requires consideration of possible changes to the original drainage channel plan as well as a full verification of the temple composition, the issue will not be considered here. It will be left for a future report.

As described here, we demonstrated the relationship between the drainage capacity of the drainage channels and observed foundation settlement in each zone. The work described here is a limited verification in which the drainage capacity of each channel is calculated assuming normal functioning, but it does confirm that the lower the maximum flow capacity of a drainage channel, the greater the foundation settlement in the vicinity. Other factors related to the foundation settlement are assumed to be the insufficient bearing capacity, deterioration of materials, opening of joints between stones owing to root growth, inhomogeneous soil forming the foundation ground, and fluctuations in the groundwater level.[Note 9)] It is inferred that these factors act together in a complex manner to produce the observed settlement. The characteristics of the south and north tower foundations are quite similar, including their scale, composition, materials, time of construction, and superstructure supported, and similar subsidence is observed at almost line-symmetrical positions. This seems to be closely related to the lack of drainage capacity of the nearby drainage channel. This leads us to inferring the following collapse mechanism for the foundation. The paving stones of each terrace foundation have a water-shedding gradient toward the respective drainage channel (Figure 6) and rainwater collects near the drainage channel. In the vicinity of the drainage channels that do not drain or that have extremely low drainage capacity, pooling rainwater seems to have infiltrated the fill within the foundation. This causes the rammed earth to flow out, reducing the volume inside the foundation and causing subsidence. As the foundation subsided, the individual building foundations above it may have also subsided.

Conclusion

Pre Rup temple (built in 961) has suffered serious foundation settlement around its outer edges and there has been much debate on the cause of this collapse. In this study, we focused on the effect of rainwater falling on the paving of the foundation, and considered the relationship between the drainage capacity and foundation settlement observed at the temple.

Pre Rup is a pyramid-type complex in which buildings have been constructed on stacked foundations. It consists of three zones. The outer first and second zones each have a surrounding wall, and each wall has eight drainage channels passing through it to discharge water into the zone below. The shape of this pipe structure, which discharges runoff water through the mouth of decorative Macara sculptures set into the terrace foundation sidewall, differs between the east-facing front of the temple and the west-facing back of the temple. It is apparent that drainage was planned considering how water would flow away to the front of the temple and for the rational discharge of rainwater at the rear of the temple. Furthermore, the relative level of these Macara water outlet spouts differs in the second and first zones, and the profile of the culvert leading to them within the terrace foundation also differs. It appears that the drainage channels of the second zone were planned to promote drainage, whereas in the first zone, a change of plan during construction resulted in a feature that promotes the retention of water. In the first zones, it seems that the drainage plan changed at some point, as evidenced by the presence of non-penetrating drainage channels, perhaps due to the interruption of the drainage channel laying work.

Given the different profiles of the drainage channels in various locations, the maximum flow capacity and rainwater runoff received by each was calculated, allowing an evaluation of the drainage capacity by comparing the two figures. The actual flows of the rainwater runoff into the drainage channels are not evaluated here; rather, the amount of runoff from each zone is evenly distributed among the channels. This causes expected errors, but does not have a significant impact on the overall drainage capacity of each zone. Drainage channels in the western half of the temple have higher maximum flow capacities than those in the eastern half, as described in the drainage plan outlined in Chapter 2. Only two drainage channels in the western half of the second zone have sufficient capacity, demonstrating that the entire zone floods during heavy rains. However, the amount of flooding is significantly greater in the first zone because of its large catchment area and the large amount of inflow from other zones as well as the fact that some of the drainage channels do not penetrate the surrounding wall.

The settlement of foundations in each zone was calculated. In the third (inner) zone, which has a three-tiered foundation, there is no major unevenness in the paving stones in the central areas of the upper two tiers. However, the curbs at the corners of each tier and in the middle of each side have subsided, whereas the central area of the first tier has also settled. In the second zone, there is little unevenness at the footings of the foundation of the upper tier and the paving stones around it, but there has been settlement of the storehouse foundations on the side nearest the drainage channels. In the first zone, there is no unevenness at the northeast foundation, but the southwest foundation has subsided by a large amount. The foundations of the eastern tower group in this zone have a large amount of subsidence, and the subsidence is almost symmetrical in the south and north tower foundations about the east-west axis. This clarifies that foundation settlement occurred in places where there were drainage channels, and points to a need to examine the relationship with rainwater drainage. Consequently, the relationship between the capacity of each drainage channel and foundation settlement was evaluated next. The factors causing subsidence are known to be diverse and complex. However, similar subsidence is observed at axisymmetric positions in the foundations of multiple towers under similar conditions, thus, it is clear that the main factor affecting subsidence at the temple is the lack of drainage capacity, particularly among the drainage channels located at the front (eastern side). It is revealed that foundation settlement is greater near the drainage channels where the maximum flow capacity is small.

Through this analysis of the Pre Rup temple, based on the analysis of the normal functioning of the drainage channels, the relationship between the drainage capacity and foundation settlement is demonstrated with some certainty. Foundation settlement resulting from poor rainwater drainage does not become apparent over a short period, but is an important issue in the long-term preservation of the temples of Angkor. The basic data presented in this paper can be considered a starting point for future efforts to conserve this historical architecture. In the particular case of Pre Rup, it is clear that the structure of the foundation was determined according to the expected magnitude of the upper load, whereas the flow capacity, and type of flow in each drainage channel was planned considering the amount of rainwater runoff expected from the paved area. Given that there was clearly such a drainage plan during construction, it is difficult to regard the inadequate drainage capacity observed in the first zone as a mere technical problem.

We point out the possibility that the drainage plan for the first zone changed after the construction of the temple, but this evaluation did not consider the overall rainwater runoff plan of the temple, including how runoff water is guided to each drainage channel; additionally, we did not attempt to consider that such a change to the drainage plan may have been made. The towers and storehouses are thought to have been added in later times and it is possible that this caused a change in the drainage channel plan.

In an upcoming paper on Pre Rup, we will analyze the water-shedding gradient and drainage gutters arrangement in the paving stones of the foundations, focusing on how water is guided to each drainage channel and documenting the drainage system of the entire temple, including any changes to the drainage plan.

To preserve historical architecture at Angkor over the long term, it is necessary to supplement any lack of drainage capacity while also maintaining the original drainage channels. Our future work will be an examination of the technical issues surrounding the measures for preserving the remaining structures, such as the installation of new drainage channels.

Acknowledgments

This work was supported by Obayashi Foundation and JSPS KAKENHI Grant Number JP 17K14797. The authors express their sincere gratitude for the help with the field survey provided by the Angkor Site International Survey Team (leader: Yoshiaki Ishizawa, Professor of Sophia University) and the APSARA National Authority(the Authority for the Protection and Safeguarding of Angkor and the Region of Angkor), former undergraduate students of the Faculty of Industrial Science and Technology, Nihon University (Toru Konishi, Takaaki Hatori, Miyuki Nagashima and Kokichi Yoshioka).

Disclosure

The authors have no conflict of interest.

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.

Notes:

^{Note 1)}Pre Rup is the name given to this temple complex in the early modern period, meaning "turning the body". The official name of the site is Rajendravadoshvara, the name of the main temple god. In an inscription (K806), it is said that King Rajendravalman ordered it built by architect Kavindrari-mathana. (References 4, 5, and 6).

^{Note 2)}The interior profile of each drainage channel was surveyed using a laser distance meter (Leica DISTO D510), and each set of data was plotted using a CAD system to create an accurate cross-sectional view.

^{Note 3)}The level of the upper surface of each foundation was obtained using a digital auto level. For the positional relationship of each foundation in the temple, the data published in Reference 3 is used.

^{Note 4)}The names and building numbers used in this paper are based on those used on the site layout drawn up by the Ecole Francaise d'Extreme-Orient (EFEO).

^{Note 5)}From Bernoulli's principle.

^{Note 6)}Areas that have undergone restoration were extracted from the annual reports of the EFEO and the reports of the Technical Committee of the International Coordinating Committee for the Safeguarding and Development of the Historic Site of Angkor (ICC) Conference (Reference 18). The ICC Conference Technical Committee has met 29 times since its first meeting (March 21, 1994).

^{Note 7)}Henri Marchal and Georges Trouve of EFEO completed the maintenance of the entire Pre Rup temple by about 1935. The photos in the report show almost the entire area of the temple.

^{Note 8)}It is assumed that the paving stones were laid on the second zone foundation after the construction of the third stage foundation and the building, thus, the level of the paving stones may have been different on each side.

^{Note 9)}In a 1930 photograph, subsidence is already visible on the southern wall and the tower. It is therefore unlikely that the main cause of the subsidence here is fluctuations in the groundwater level resulting from increased tourism in recent years, which has increased the pumping of groundwater in the area around the Angkor monuments.